pytorch

mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 12:54:11 +08:00

Files

Nicolas De Carli eb11d172e3 [Caffe2] Improve SVE batch box cox by 2% (#163360 )

Summary:
Improve bound checking on exp computation, decreasing the longest dependency chain by 1.

Box-cox benchmarks show about 2% of improved throughput.
Precision remains unaltered.

before:

NonZeroLambdaBatch                                        155.30us     6.44K

after:

NonZeroLambdaBatch                                        151.78us     6.59K

Test Plan:
Correctness:

buck2 test @//mode/opt //koski/functions_contrib/df4ai/tests:batch_box_cox_test

Performance:

buck2 run @//mode/opt //koski/functions_contrib/df4ai/benchmark:boxcox_benchmark

Differential Revision:
D82847111

Privacy Context Container: L1208939

Pull Request resolved: https://github.com/pytorch/pytorch/pull/163360
Approved by: https://github.com/Skylion007

2025-09-20 06:42:26 +00:00

batch_box_cox_avx512.cc

[caffe2] Add AVX512 support for box_cox operator (#143627 )

2025-01-07 09:54:39 +00:00

batch_box_cox_sve128.cc

[Caffe2] Improve SVE batch box cox by 2% (#163360 )

2025-09-20 06:42:26 +00:00

batch_box_cox_vec.h

Revert "Fix boxcox to return same result for same input in one batch (#162772 )"

2025-09-19 17:58:29 +00:00

CMakeLists.txt

Add SVE implementation of embedding_lookup_idx (#133995 )